Model Refactor (#66)

* initial implementation of pipeline wrapper and model updates

* merge model card and config, update docs

* update examples readme

* review comments and model cleanup

* expose a better neighbor list mechanism and fix some aimnet + pipeline issues, add example

* make some examples run and update aimnet to use neighbors utilities

* update aimnet, other review comments

* more review responses and improve aimnet coverage

* preempt merge conflict in 01 example

* bug in pipeline and aimnet2

* more merge conflict resolutions

* fix model_config mutation in pipeline

* fix missing skin argument

* remove errant neighbor list file

* add packages and fix mace test bug

* fix skin dropping

Author: dallasfoster

docs/conf.py

@@ -74,7 +74,6 @@
     "sphinx_togglebutton",
     "sphinx.ext.graphviz",
     "sphinx_gallery.gen_gallery",
-    "model_card_ext",
 ]
 
 source_suffix = [".rst", ".md"]
@@ -86,7 +85,6 @@
 exclude_patterns = [
     "_build",
     "sphinxext.py",
-    "model_card_ext.py",
     "Thumbs.db",
     ".DS_Store",
 ]

docs/index.md

@@ -46,7 +46,7 @@ batching and size-aware sampling handle load balancing automatically.
 ## Highlights
 
 - **Bring your own model** --- wrap MACE, AIMNet2, or any PyTorch MLIP in a
-  few lines with a standardized `ModelCard` interface.
+  few lines with a standardized `ModelConfig` interface.
 - **Compose, don't configure** --- fuse stages on one GPU with `+`, distribute
   across GPUs with `|`, and inject behavior at nine hook points per step.
 - **GPU-native data** --- `AtomicData` and `Batch` are Pydantic-validated,